Technical description, phantom accuracy, and clinical feasibility for fiducial-free frameless real-time image-guided spinal radiosurgery

Restricted access


The authors describe the technical application of the Xsight Spine Tracking System, data pertaining to accuracy obtained during phantom testing, and the initial clinical feasibility of using this fiducial-free alignment system with the CyberKnife in spinal radiosurgery.


The Xsight integrates with the CyberKnife radiosurgery system to eliminate the need for implantation of radiographic markers or fiducials prior to spinal radiosurgery. It locates and tracks spinal lesions relative to spinal osseous landmarks. The authors performed 10 end-to-end tests of accuracy using an anthropomorphic head and cervical spine phantom. Xsight was also used in the treatment of 50 spinal lesions in 42 patients. Dose planning was based on 1.5-mm-thick computed tomography slices in which an inverse treatment planning technique was used.

All lesions could be treated using the fiducial-free tracking procedure. Phantom tests produced an overall mean targeting error of 0.52 ± 0.22 mm. The setup time for patient alignment averaged 6 minutes (range 2–45 minutes). The treatment doses varied from 12 to 25 Gy to the median prescription isodose of 65% (40 to 70%). The tumor volume ranged between 1.3 and 152.8 cm3The mean spinal cord volume receiving greater than 8 Gy was 0.69 ± 0.35 cm3No short-term adverse events were noted during the 1- to 7-month follow-up period. Axial and radicular pain was relieved in 14 of 15 patients treated for pain.


Fiducial-free tracking is a feasible, accurate, and reliable tool for radiosurgery of the entire spine. By eliminating the need for fiducial implantation, the Xsight system offers patients noninvasive radiosurgical intervention for intra- and paraspinal tumors.

Abbreviations used in this paper:CT = computed tomography; DRR = digitally reconstructed radiograph; EBRT = external-beam radiotherapy; LINAC = linear accelerator; MR = magnetic resonance; ROI = region of interest; 2D = two-dimensional; 3D = three-dimensional.

Article Information

Address reprint requests to: Alexander Muacevic, M.D., European CyberKnife Center Munich, Max-Lebsche-Platz 31, 81377 Munich, Germany. email:

© AANS, except where prohibited by US copyright law.



  • View in gallery

    Results of the image-enhancement process of the DRR (A) and the intratreatment radiograph (B). Both sets of radiographs undergo image processing based on top-hat filtering to improve the visualization of the osseous anatomy (C and D). These images were obtained in the patient in the Illustrative Case.

  • View in gallery

    Image ROI selection for spinal tracking. An ROI containing the maximum osseous anatomical information surrounding the target volume is selected based on an initial user-defined position, which is refined by an algorithm that seeks to maximize the image entropy within the ROI. The resulting ROI typically includes two vertebral bodies, which form the basis of the treatment alignment.

  • View in gallery

    The hierarchical mesh tracking procedure. Registration is performed using a hierarchical mesh technique in which the calculation is made at a series of discrete points within the ROI. The process is iterative, with additional registration points added at each step to improve the spatial resolution of the result. This approach generates a deformable registration model, which can account for nonrigid changes in the patient posture between pre- and intratreatment imaging. The deformation is apparent in the mesh in the x-ray panel.

  • View in gallery

    Photograph of patient position during spinal radiosurgery for a lumbar lesion. The patient lies on the treatment couch without any vacuum bags or fixation devices. A cushion is placed under the legs for comfort. The pretreatment CT scan was obtained with the patient in the same position.

  • View in gallery

    Left: Sagittal T2-weighted MR image obtained in a patient undergoing surgery and conventional radiotherapy for a malignant peripheral nerve sheath tumor at the C-6 level (arrow). The recurrent tumor was compressing the spinal cord anteriorly. Instead of repeated surgery, CyberKnife radiosurgery was performed using fiducial-free tracking. Right: Sagittal image acquired after spinal radiosurgery. The tumor has shrunk significantly (arrow) and the anterior cord compression has been eliminated.

  • View in gallery

    Axial (A), coronal (B), and sagittal (C) views of a treatment plan. Note the sharp dose gradient near the spinal cord. The thick green line represents the prescribed isodose, and the blue and pink shaded regions adjacent to the treated volume represent dose constraints placed around the spinal cord.

  • View in gallery

    Cumulative dose–volume histogram showing the planning target volume (PTV) and spinal cord coverage. The spinal cord was completely spared at the 80% dose prescription level.

  • View in gallery

    Three-dimensional translational motion in millimeters (upper) and the individual components (rotation angles) of orientation in degrees (lower), plotted as a function of treatment node during the first path of treatment. The patient’s position and orientation are recorded every three nodes. At Node 22 (~ 17 minutes into the procedure), because the patient’s movements caused a left–right roll error, an automatic couch correction was necessary. The smaller corrections (≤1.8 mm) until the end of the path were made automatically by the robot. CCW/CW = counterclockwise/clockwise.



Adler JR JrChang SDMurphy MJDoty JGeis PHancock SL: The Cyberknife: a frameless robotic system for radiosurgery. Stereotact Funct Neurosurg 69:1241281997


Degen JWGagnon GJVoyadzis JMMcRae DALunsden MDieterich S: CyberKnife stereotactic radiosurgical treatment of spinal tumors for pain control and quality of life. J Neurosurg Spine 2:5405492005


Fu DKuduvalli GEnhancing skeletal features in digitally reconstructed radiographs. inReinhardt JMPluim JP: (eds)Medical Imaging 2006: Image Processing San DiegoThe International Society for Optical Engineering2006. Vol 6144:(Abstract 61442M)


Fu DKuduvalli GMaurer CR JrAllision JWAdler JR Jr: 3D target localization using 2D local displacements of skeletal structures in orthogonal x-ray images for image-guided spinal radiosurgery. Int J CARS 1:Suppl 11982002006


Gerszten PCBurton SAOzhasoglu CVogel WJQuinn AECyberknife radiosurgery: single-fraction treatment for spinal tumors. inMould RF: (ed)Robotic Radiosurgery Sunnyvale, CACyberKnife Society Press2005. Vol 1:pp171186


Gerszten PCGermanwala ABurton SAWelch WCOzhasoglu CVogel WJ: Combination kyphoplasty and spinal radiosurgery: a new treatment paradigm for pathological fractures. J Neurosurg Spine 3:2963012005


Gerszten PCOzhasoglu CBurton SAVogel WAtkins BKalnicki S: Evaluation of CyberKnife frameless real-time image-guided stereotactic radiosurgery for spinal lesions. Stereotact Funct Neurosurg 81:84892003


Gerszten PCOzhasoglu CBurton SAVogel WJAtkins BAKalnicki S: CyberKnife frameless stereotactic radiosurgery for spinal lesions: clinical experience in 125 cases. Neurosurgery 55:89992004


Gerszten PCOzhasoglu CBurton SAWelch WCVogel WJAtkins BA: CyberKnife frameless single-fraction stereotactic radiosurgery for tumors of the sacrum. Neurosurg Focus 15:2E72003


Kuo JSYu CPetrovich ZApuzzo ML: The CyberKnife stereotactic radiosurgery system: description, installation, and an initial evaluation of use and functionality. Neurosurgery 53:123512392003


Murphy MJChang SDGibbs ICLe QTHai JKim D: Patterns of patient movement during frameless image-guided radiosurgery. Int J Radiat Oncol Biol Phys 55:140014082003


Ryu SIChang SDKim DHMurphy MJLe QTMartin DP: Image-guided hypo-fractionated stereotactic radiosurgery to spinal lesions. Neurosurgery 49:8388462001


Welch WCGerszten PC: Accuray CyberKnife image-guided radiosurgical system. Expert Rev Med Devices 2:1411472005


Yu CMain WTaylor DKuduvalli GApuzzo MLAdler JR Jr: An anthropomorphic phantom study of the accuracy of Cyberknife spinal radiosurgery. Neurosurgery 55:113811492004




All Time Past Year Past 30 Days
Abstract Views 52 52 25
Full Text Views 54 54 12
PDF Downloads 23 23 6
EPUB Downloads 0 0 0


Google Scholar